Market Risk Insights

Problem

• Market risk judgments are scattered across quotes, valuations, funding flows, and macro sentiment. Looking at one indicator in isolation can be misleading, and manually compiling cross-market data with source tracking is time-consuming.

How It Works

• The skill runs a two-stage workflow on the Hengsheng Juyuan MCP financial database: Phase 1 data collection and Phase 2 risk analysis.
• It covers equity, bond, commodity, and overseas markets, collecting index prices, valuation percentiles, margin trading, northbound flows, credit spreads, commodity sectors, FX, US Treasuries, and industry rotation.
• Phase 1 only collects data and records timestamps and sources; failed points are retried once and then marked null without blocking the pipeline.
• Phase 2 applies rule-based and composite scoring to assign low, medium, or high risk levels, with upgrades for persistent main outflows, rapid margin balance declines, and sustained northbound outflows.
• It outputs MD, HTML, and PDF reports, generates a risk heatmap, and enforces one-to-one footnote mapping to the data source table.

Boundaries

• Best for daily risk snapshots, allocation context, and anomaly monitoring; not for real-time trade signals or single-stock diagnosis.
• Results depend on the quality of MCP API data; missing values are marked null rather than invented.
• Risk levels are rule-based judgments and should be used within your own investment constraints and compliance process.